parsing a string in SAS 9.2 that contains the '|' character - regex

Question

I have a variable that contains a number of firms separated by the | symbol. I would like to be able to count how many firms there. i.e., the number of | + 1, and ideally identify the location of the | symbol in the string. Note there will not be more than five firms in a single variable. I was trying to use the following approach but run into the fact that SAS treats the | symbol as a special operator.
pattern1 = prxparse('/|/'); /* I can't seem to get SAS to treat this as a text to compare */
start = 1;
stop = length(reassignment2); /* my list of firms is in the variable reassignment2 */
call prxnext(pattern1, start, stop, reassignment2, position, length);
ARRAY Y[5];
do J=1 to 5 while (position > 0);
Y[J]=position;
call prxnext(pattern1, start, stop, reassignment2, position, length);
end;
nfirms=j+1;
run;

Answer 1

I would do it somewhat differently. What you really want is not the number of | characters, but the actual firms, right? So search for those. Your code had a number of minor issues; primarily, you must first prxmatch before using call prxnext, your j+1 is wrong because the loop iterator actually increments one beyond the last qualifying loop value (I use j-1 because I will find one more element than you), and | is a regular expression metacharacter and must be escaped if you actually want to use it, unless it is inside [] like I am using it.
data test;
infile datalines truncover;
input #1 reassignment2 $50.;
pattern1 = prxparse('/[^|]+/io'); /* Look for non-| characters */
start = 1;
stop = length(reassignment2); /* my list of firms is in the variable reassignment2 */
rc=prxmatch(pattern1,reassignment2);
if rc>0 then do;
ARRAY Y[5];
do J=1 by 1 until (position = 0);
call prxnext(pattern1, start, stop, reassignment2, position, length);
Y[J]=position;
end;
nfirms=j-1;
end;
else nfirms=0;
put nfirms=;
datalines;
Firm1|Firm2|Firm3
Firm1|Firm2|Firm3|Firm4
Firm1
Firm1|Firm2
;;;;
run;
For completeness' sake, you could also do this easily without regular expressions, using call scan.
data test;
infile datalines truncover;
input #1 reassignment2 $50.;
array y[5];
do nfirms=1 by 1 until (position le 0);
call scan(reassignment2,nfirms,position,length,'|');
y[nfirms]=position;
end;
nfirms=nfirms-1; *loop ends one iteration too late;
put nfirms=;
datalines;
Firm1|Firm2|Firm3
Firm1|Firm2|Firm3|Firm4
Firm1
Firm1|Firm2
;;;;
run;

Answer 2

I agree with #Joe that this could be done more simply without regular expressions, though I would simplify his code a little further to exclude the use of an array.
data test;
infile datalines truncover length = reclen;
input firmlist $varying256. reclen;
i = 0;
do until(scan(firmlist,i,"|") = "");
i + 1;
end;
nfirms = i - 1;
drop i;
datalines;
Firm1|Firm2|Firm3
Firm1|Firm2|Firm3|Firm4
Firm1
Firm1|Firm2
;
run;
You said you'd also like to capture the position of the "|" character in the string, but if there are multiple firms per record there will be multiple "|" characters in the string. If you want the position of each one, an array might be a better route, though if you only want one, the index function will get you what you want. You'd use delimpos = index(firmlist,"|");.
I hope that helps!